# WaveFit: An Iterative and Non-autoregressive Neural Vocoder based on Fixed-Point Iteration

Source: [WaveFit: An Iterative and Non-autoregressive Neural Vocoder based on Fixed-Point Iteration](<https://www.sota2.com/research/paper/68fa6d901e209131cea3aa80>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-26T00:37:22.464Z

Authors: Yuma Koizumi, Kohei Yatabe, Heiga Zen, Michiel Bacchiani

Published: 2022-10-03T15:45:05.000Z

## Abstract

Denoising diffusion probabilistic models (DDPMs) and generative adversarial networks (GANs) are popular generative models for neural vocoders. The DDPMs and GANs can be characterized by the iterative denoising framework and adversarial training, respectively. This study proposes a fast and high-quality neural vocoder called \\textit{WaveFit}, which integrates the essence of GANs into a DDPM-like iterative framework based on fixed-point iteration. WaveFit iteratively denoises an input signal, and trains a deep neural network (DNN) for minimizing an adversarial loss calculated from intermediate outputs at all iterations. Subjective (side-by-side) listening tests showed no statistically significant differences in naturalness between human natural speech and those synthesized by WaveFit with five iterations. Furthermore, the inference speed of WaveFit was more than 240 times faster than WaveRNN. Audio demos are available at \\url{google.github.io/df-conformer/wavefit/}.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Neural Vocoding on LibriTTS-R clean-100 (test)](<https://www.sota2.com/research/sota/neural-vocoding-on-libritts-r-clean-100-test.md>): Neural Vocoding; LibriTTS-R clean-100 (test); NISQA: 4.02.

- [Waveform Generation on LibriTTS-R clean (test)](<https://www.sota2.com/research/sota/waveform-generation-on-libritts-r-clean-test.md>): Waveform Generation; LibriTTS-R clean (test); Speech BERT Score: 94.28.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2210.01029>)

[PDF](<https://arxiv.org/pdf/2210.01029.pdf>)
